Breast cancer is one of the most malignant and dangerous illnesses that cause death for women in the world. Early classification of breast cancer plays a vital role in saving the lives of many sick women. The spread of breast cancer has increased the diagnostic burden for doctors and radiologists so they turned to using Computer-Aided Diagnosis. This paper proposed Machine learning and Deep learning for smart diagnosis of breast cancer based on thermal images. The proposed machine learning algorithm is investigated by using three stages, which are preprocessing, feature extraction and Artificial Neural Network classifier. Also, four Deep learning models AlexNet, GoogleNet, SqueezNet and ResNet18 are proposed to achieve accurate diagnosis of breast cancer and achieve high performance. In simulation results, the Machine learning algorithm achieves the accuracy of 84,62 %, sensitivity =76,92 %, and specificity =88,46% In addition Deep learning models achieve accuracy arrive to 100% especially AlexNet.